Aladin consortium activities in data assimilation

Dubrovnik October 8-11, 2007 EWGLAM/SRNWP meetings

Aladin consortium activities in Aladin consortium activities in data assimilationdata assimilation

Claude Fischer & Gergely BölöniClaude Fischer & Gergely Bölöni

Stolen material from a real big number of really Stolen material from a real big number of really nice collaboratorsnice collaborators


Overview Overview

1.1. Mainframe 3D-VAR applicationMainframe 3D-VAR application

2.2. Background error statistics: « B » Background error statistics: « B » matrixmatrix

3.3. Observations and OSEObservations and OSE

4.4. Surface assimilation => talk by JF Surface assimilation => talk by JF MahfoufMahfouf

5.5. Plans and collaborationsPlans and collaborations


Mainframe 3D-VAR in AladinMainframe 3D-VAR in Aladin

►Example of Aladin-FranceExample of Aladin-France►Preparations at CHMIPreparations at CHMI►Aladin Rapid Update CycleAladin Rapid Update Cycle►Digital filter blendingDigital filter blending►3D-VAR Aladin coupled with IFS LBC3D-VAR Aladin coupled with IFS LBC


Mainframe 3D-VAR: example of Mainframe 3D-VAR: example of Aladin-FranceAladin-France

► Incremental 3D-VARIncremental 3D-VAR► Continuous assimilation cycle, 6 hour frequency, long cut-off Continuous assimilation cycle, 6 hour frequency, long cut-off

assimilation cycle and short cut-off production, coupled with assimilation cycle and short cut-off production, coupled with Arpège, Analysis=Model gridmeshArpège, Analysis=Model gridmesh=9.5km=9.5km

► Observations:Observations: SHIP winds, SHIP winds, synop & radome Ps, T2m, RH2m, V10msynop & radome Ps, T2m, RH2m, V10m Aircraft dataAircraft data SATOB motion windsSATOB motion winds Drifting buoysDrifting buoys Soundings (TEMP, PILOT)Soundings (TEMP, PILOT) Satellite radiances: AMSU-A, MHS, HIRS (NOAA & METOP), Satellite radiances: AMSU-A, MHS, HIRS (NOAA & METOP),

Meteosat-9 SEVIRIMeteosat-9 SEVIRI QuikSCAT windsQuikSCAT winds Ground-based GPS zenital delaysGround-based GPS zenital delays

► Digital filter initialisation (non incremental)Digital filter initialisation (non incremental)


Implementation of ALADIN 3DVAR Implementation of ALADIN 3DVAR at CHMI at CHMI ((A. Trojakova)A. Trojakova)

Present status:• operational surface OI (together with a DF blending cycle)• use of ODB (SYNOP and TEMP data)

News:• B matrix computed for the CZ domain (standard and lagged NMC for Autumn 2006)• validation of 3DVAR is on the way (CY32)• observation preprocessing further developed for SEVIRI (phased to CY32)

standard

NMC

lagged

NMC


Blending by Digital Filter (Blending by Digital Filter (M. BelluM. Belluss, , M. Derkova)M. Derkova)

• Blending by DF implemented in ALADIN/SHMU

• DF settings tuned for the SHMU domain

• DF Blending cycle in parallel suite since the 6th of

August 2007 => operationally implemented since September 19th, 2007

• case studies and subjective evaluation good results (localization of precipitation)


Blending by Digital Filter

verif

blendingoper

Case study 11/08/2007

Better localization of the

precipitation in the DF blending

suite


ALADIN Rapid Update Cycle (ALADIN Rapid Update Cycle (G. G. Bölöni, S. Kertesz, B. Strajnar)Bölöni, S. Kertesz, B. Strajnar)

• 3 and 1 hourly RUC compared to the usual 6h cycle • cycle setup changed for RUC (LBC and obs extraction)• expected extra from RUC:

more SYNOPs (only Ps)

smaller error in the innovation vector

(due to more frequent analysis)

• preliminary results from parallel suites with RUC 3h are presented

results depend a lot on the obs usage of

the reference

ARPE: 6h cycle using an operational cut-off, i.e. +/-3h

(all aircrafts and satellites)

RUC6: 6h cycle using the same cut-off as RUC3, i.e. +/-1.5 h

(some aircrafts and satellites excluded)


ALADIN Rapid Update Cycle

Reference: ARPE

U

RHU

Red: improvement for U

Result as one would expect…

(…strictly the impact of more SYNOPs and a more exact innovation vector…)

Blue: degradation for RHU

Deficiency in the 3h background forecast (non-appropriate Jb)?


ALADIN 3DVAR cycle coupled with ALADIN 3DVAR cycle coupled with ECMWF LBC (ECMWF LBC (S. Kertesz)S. Kertesz)

• Dynamical adaptation and 3DVAR assimilation cycle coupled with ECMWF LBC• operational constraint: at 00UTC ECMWF LBCs from the previous 18UTC run are

available

Conclusions:

Surface initial field should be ARPEGE or local

analysis (ECMWF surface inconsistent with ISBA)

Dynamical adaptation: coupling with ARPEGE

00UTC is better than coupling with ECMWF 18UTC

Assimilation: coupling with ECMWF 18UTC run is

better than coupling with ARPEGE 00UTC run (slight

improvement for the upper air, similar results on the surface)

The 6h shift in the LBC has a very large impact i.e. coupling

with 00UTC ECMWF is much better than coupling with

ARPEGE 00UTC both for Dyn. Adap and 3DVAR assimilation


ALADIN 3DVAR cycle coupled with ECMWF LBC

Verification against radiosondes

GEO

U

Dyn. Ad Assim

LBC: ECMWF 18UTC – ARPEGE 00UTC

(taking into account the

operational constraint)

Red: ECMWF better than

ARPEGE


« B matrix » evolutions« B matrix » evolutions

►Structure functions: new humidity Structure functions: new humidity controlcontrol

►Sampling data for statistics: Sampling data for statistics: ensemblesensembles

►A posteriori tuning: … A posteriori tuning: …


ContextContext►Following the work by Elias Holm at ECMWF Following the work by Elias Holm at ECMWF ►Humidity seems to have the least Gaussian Humidity seems to have the least Gaussian

and the least homogeneous background and the least homogeneous background errors of all analysis variables.errors of all analysis variables.

► Improving the description of the humidity Improving the description of the humidity background errors may improve the background errors may improve the humidity analysis.humidity analysis.

►This can be done by changing the current This can be done by changing the current humidity control variable to one in which the humidity control variable to one in which the errors are more Gaussian and homogeneous.errors are more Gaussian and homogeneous.

The new formulation of the Humidity control The new formulation of the Humidity control variable in Aladin (L. Berre, R. El Ouaraini)variable in Aladin (L. Berre, R. El Ouaraini)


The new humidity control variableThe new humidity control variable

We change δq (specific humidity) by the normalized relative humidity We change δq (specific humidity) by the normalized relative humidity δδRH_nRH_n::

δδRH_n = RH_n = δδRHRH / / σσ(RH(RHbb+δ+δRH/2)RH/2)

P.S.:P.S.:► The standard deviation The standard deviation σσ depends depends

on the humidity guess value RHon the humidity guess value RHbb..► This new formulation avoids negativeThis new formulation avoids negative

and supersatured humidity and supersatured humidity

(Smaller(Smaller σ σbb near 100%near 100% and 0% : and 0% : see figure)see figure)


Some resultsSome results

The new humidity control variable (Experiment) The current humidity control variable (Reference)

•Both analyzes lead to a dryer analyzed state than the first guess (in blue) with only local moistening (of smaller amplitude).

•The drying is however less pronounced with the new control (maximum of –39 % against –63 %).

•The spatial patterns of drying are wider spread and more homogeneous with the new control.


Ensemble B Ensemble B at HMS in operational 3D-at HMS in operational 3D-VARVAR ( (G. Bölöni)G. Bölöni)

• Arpege ensemble downscaled with CY30 (new ensemble B + tuned sigmab’s [Désroziers’ a posteriori tuning])• 1 month parallel suite (comparison with the operational [NMC])• good scores ensemble B in operation since end of August

RMSE vertical cross-sections:- 00 UTC run- comparison against ECMWF analysis- reddish: ensemble better than NMC

GEO T RHU


Observations & O.S.E.Observations & O.S.E.

►Surface emissivities, surface Surface emissivities, surface temperature and radiances over landtemperature and radiances over land

►Radar windsRadar winds►Radar reflectivitiesRadar reflectivities


Land Surface Emissivity at Microwave Land Surface Emissivity at Microwave Frequencies For Satellite Data Assimilation (F. Frequencies For Satellite Data Assimilation (F.

Karbou)Karbou)Over land:

The surface emissivity is higher (~1.0) than over sea (~0.5) The surface emissivity is higher (~1.0) than over sea (~0.5) the the surface contribution to the measured radiance is largersurface contribution to the measured radiance is larger

The land emissivity varies at least with surface condition, roughness, The land emissivity varies at least with surface condition, roughness, moisture, …moisture, …

Residual uncertainties about land emissivity and skin temperature Residual uncertainties about land emissivity and skin temperature Only microwave channels that are the least sensitive to the surface Only microwave channels that are the least sensitive to the surface are assimilatedare assimilated

Existing emissivity models : facilitate the assimilation of channels that Existing emissivity models : facilitate the assimilation of channels that receive a contribution from the surface receive a contribution from the surface

BUT…..BUT….. The models need accurate input parameters hardly available at The models need accurate input parameters hardly available at global scaleglobal scale

Need for alternatives to estimate the land emissivity Need for alternatives to estimate the land emissivity


Three land surface parameterizations with increasing complexity have been Three land surface parameterizations with increasing complexity have been tested using AMSU-A, MHS (AMSU-B) and SSM/I observations (Karbou et tested using AMSU-A, MHS (AMSU-B) and SSM/I observations (Karbou et al. 2006):al. 2006):

EXP_ATLAS:EXP_ATLAS: Averaged emissivities over 2 weeks prior to the Averaged emissivities over 2 weeks prior to the assimilation period; Ts is taken from the model’ FG.assimilation period; Ts is taken from the model’ FG.

EXP_DYN:EXP_DYN: Dynamically varying emissivities derived at each pixel Dynamically varying emissivities derived at each pixel using only one channel of each instrument; Ts is taken from the using only one channel of each instrument; Ts is taken from the model FG.model FG.

EXP_SKIN:EXP_SKIN: Averaged emissivities + dynamically estimated skin Averaged emissivities + dynamically estimated skin temperature Ts at each pixel using one (or two) channel(s) of each temperature Ts at each pixel using one (or two) channel(s) of each instrument instrument

All surface parameterizations are handled by the RTTOV model (Eyre All surface parameterizations are handled by the RTTOV model (Eyre 1991; Saunders et al. 1999; Matricardi et al. 2004)1991; Saunders et al. 1999; Matricardi et al. 2004)

The land surface emissivity methods have also been tested within the IFS The land surface emissivity methods have also been tested within the IFS system and have been adapted to SSMI/S, TMI, AMSR-E observations in system and have been adapted to SSMI/S, TMI, AMSR-E observations in addition to AMSU, SSM/I (Karbou et al. 2007, SAF/NWP Report)addition to AMSU, SSM/I (Karbou et al. 2007, SAF/NWP Report)

Karbou, F., E. Gérard, and F. Rabier, 2006, Karbou, F., E. Gérard, and F. Rabier, 2006, Microwave Land Emissivity and Skin Temperature for AMSU-A & -B Assimilation Over Land, Microwave Land Emissivity and Skin Temperature for AMSU-A & -B Assimilation Over Land, Q. J. R. Q. J. R.

Meteorol. SocMeteorol. Soc., ., vol 132, No. 620, Part A, pp. 2333-2355(23) .vol 132, No. 620, Part A, pp. 2333-2355(23) .Karbou, F., N. Bormann, and J-N., Thépaut, 2007 , Karbou, F., N. Bormann, and J-N., Thépaut, 2007 ,

Towards the assimilation of SSMI/S observations over land, RAPPORT NWP-SAFTowards the assimilation of SSMI/S observations over land, RAPPORT NWP-SAF

OverviewOverview


Improvement in the performances of the RTTOV model: bias/std, Improvement in the performances of the RTTOV model: bias/std, increase of the number of observations that could be assimilated increase of the number of observations that could be assimilated

Forecast scores globally neutral to positive for humidity, temperature Forecast scores globally neutral to positive for humidity, temperature and geopotential height.and geopotential height.

Precipitation forecasts improved for West Africa. Further evaluation will Precipitation forecasts improved for West Africa. Further evaluation will be performed for AMMA (summer 2006) and with a limited area model for be performed for AMMA (summer 2006) and with a limited area model for intense Mediterranean events.intense Mediterranean events.

Main ResultsMain Results

Control

Clo

ud

fore

casts

(2

00

50

82

7+

18

h)

Exp_dyn: 5 SSM/I

channels over land IR- Meteosat: 20050827-18hIR- Meteosat: 20050827-18h

Ongoing experiments to better understand the impact of changes in the Ongoing experiments to better understand the impact of changes in the surface (emissivity/skin temperature), bias correction, cloud identificationsurface (emissivity/skin temperature), bias correction, cloud identification

Sensitivity studies to assimilate cloudy/rainy microwave observations Sensitivity studies to assimilate cloudy/rainy microwave observations over sea/landover sea/land


Fg-departures (obs-guess) global histograms, 15-31 August 2005Fg-departures (obs-guess) global histograms, 15-31 August 2005

50.3 GHzCh3 AMSU-A

ControlControlATLASATLAS from 50GHzfrom 50GHz

EMIS-DYNEMIS-DYN from 23GHzfrom 23GHz

ATLAS+SKINATLAS+SKIN from 50GHz, Ts from 50GHz, Ts from 23GHzfrom 23GHz

89 GHzCh15 AMSU-A

150 GHzCh2 AMSU-B

Results: Observation operator simulationsResults: Observation operator simulations


Assimilation of radar data (T. Assimilation of radar data (T. Montmerle & C. Faccani)Montmerle & C. Faccani)

The ARAMIS radar networkCurrent status : 16 Doppler C-band Radars performing

between 2 and 11 PPIs / 15’• 1 double polarimetric (Trappes)

Assimilation of reflectivities

Tested in AROME, still technical developments to perform

Assimilation of Doppler windsTested with ALADIN, currently running in pre-

operational mode in AROME

Future Doppler Radar

Doppler Radar

Future Radar with double polarisation


Without Radar 72LM

Analysis of divergence(z=2500m)

20070525 - 15 UTC

Independent wind retrieval from multiple radar measurements (CMR / Muscat)

With Radar

Main convergence line is shifted by ~ 60 km

Assimilation of radial winds in AROME 3DVar


R-15UTC1h Cumulated rainfall 16UTC->18UTC

Without radar

With radar

AROME Spin-up

Observations(Trappes radar)

However: no real impact when the dynamical structure of convective systems is not sampled in and at the top of the boundary layer

Main convergence line is well analyzed thanks to Doppler observations

=> Squall line forecast is more realistic with a better persistence of rain


Goal Goal ::Operationnally assimilate radar reflectivities in AROME by 2009-Operationnally assimilate radar reflectivities in AROME by 2009-

20102010

StatusStatus : :► Volumic (3D) reflectivity data routinelyVolumic (3D) reflectivity data routinely available since August available since August

2007, in real time. Pre-processing check to remove erroneous 2007, in real time. Pre-processing check to remove erroneous data (soil and sea clutters, …)data (soil and sea clutters, …)

► Reflectivity observation operator readyReflectivity observation operator ready, simulates modelled , simulates modelled reflectivities.reflectivities.

► Quality control checkQuality control check by a gross comparison of observed and by a gross comparison of observed and modelled columns.modelled columns.

► Assimilation in the AROME system via a 1D+3DVarAssimilation in the AROME system via a 1D+3DVar: : reflectivities are inverted into pseudo-observations of relative reflectivities are inverted into pseudo-observations of relative humidity profiles (whose impact is expected to be bigger than humidity profiles (whose impact is expected to be bigger than when modifying the hydrometeor fields).when modifying the hydrometeor fields).

Reflectivity assimilation (E. Wattrelot, Reflectivity assimilation (E. Wattrelot, O. Caumont, M. JuraO. Caumont, M. Juraššek, G. Haase, …)ek, G. Haase, …)


21HRaw Radar Composite

00H Raw Radar Composite

21H 21H

00H 00H

REFLEC

REFERENCE

REFERENCE

REFLEC

Composite radar images (left panels), 3h forecasts of the 2500 m reflectivity simulated by the REFLEC experience (middle panels) and the REF (right panels) experiments at 21h UTC (top panels) and at 00h UTC the 2nd of August (bottom panels)

2 experiments :-REF run is an AROME 3h RUC without reflectivities (4 assimilation/forecast cycles at 18h)-REFLEC with assimilated reflectivities by 1D+3DVar method

Impact of reflectivity assimilation on the Impact of reflectivity assimilation on the case of 1st August 2007case of 1st August 2007


Top left panel: 3 h modelled precipitations (6-3 h lead times) in REFLECTop right panel: ibid without reflectivity assimBottom left panel: 3h raingauge aggregatesBottom right panel: POD v/s FAR diagram for REFLEC (red) and REF (green)

2 experiments :-REF run is an AROME 3h RUC without reflectivities (4 assimilation/forecast cycles starting 01/08/07 at 18 UTC)-REFLEC with assimilated reflectivities by 1D+3DVar method

Impact of reflectivity assimilation on the Impact of reflectivity assimilation on the case of 1st-2nd August 2007case of 1st-2nd August 2007


Future workFuture work::

► Improvement of pre-processing for erroneous cluttersImprovement of pre-processing for erroneous clutters (by data (by data providers). For instance, fix echoes can be almost perfectly providers). For instance, fix echoes can be almost perfectly eliminated on polarimetric radars. Signal damping by eliminated on polarimetric radars. Signal damping by precipitations also can be well corrected on polarimetric precipitations also can be well corrected on polarimetric radars.radars.

► Improvement on the observation operator: Improvement on the observation operator: inclusion of a inclusion of a minimum height of visibility within the width of the beamminimum height of visibility within the width of the beam (useful for the vertical interpolation of model data on the path (useful for the vertical interpolation of model data on the path of the beam)of the beam)

► Improvement of the 1D Bayesian retrieval methodImprovement of the 1D Bayesian retrieval method: ensure a : ensure a better consistency between the sets of modelled columns and better consistency between the sets of modelled columns and the observed ones (allow for « wet columns » in rainy areas the observed ones (allow for « wet columns » in rainy areas even if surrounding model has no rain and vice versa). Extend even if surrounding model has no rain and vice versa). Extend the 1D inversion to temperature, wind …the 1D inversion to temperature, wind …

► Evaluation of the 1D+3DVar method on convective cases Evaluation of the 1D+3DVar method on convective cases poorly forecast by Aromepoorly forecast by Arome

► Improvement of quality control and thinning of radar inverted Improvement of quality control and thinning of radar inverted profilesprofiles (presently relies on what is done for radiosoundings). (presently relies on what is done for radiosoundings).


Concept of visibility in the observation operator

• the visibility is the minimum height which is detectable by the radar main lobe. It depends on the elevation angle, the beam width and the topography

•Currently, topographical beam blockage is not considered. This might cause problems in mountaineous regions where the interpolation considers model levels which are not visible by the radar

• By using visibility maps for standard propagation the vertical interpolation becomes more realistic where the radar beam is partly blocked

Vertical interpolation from model to radar space (along dashed lines). The black solid line corresponds to the beam center while the dotted lines mark the beam width of the unblocked beam. The red solid line

defines the actual visibility at this elevation angle assuming atmospheric standard conditions. The circles indicate the integration

limits used in the vertical interpolation for the blocked (red) and unblocked beam (black), respectively

Topography interpolated on Bollène radar geometry (left), visibility of the lowest elevation of Bollene radar (middle) and differences of simulations with and without taking into account beam blockage information through visibility maps (right)


Surface assimilationSurface assimilation► O.I. scheme (CANARI) in Arpège and Aladin O.I. scheme (CANARI) in Arpège and Aladin

(CZ and MO; tested in HU and FR)(CZ and MO; tested in HU and FR)► Same algorithm to be ported into the Same algorithm to be ported into the

externalized surface scheme (SURFEX) externalized surface scheme (SURFEX) ► New system for 2D analysis (PBL fields and New system for 2D analysis (PBL fields and

spatialisation tool) ?spatialisation tool) ?► Development of a new 2D-VAR (aka Development of a new 2D-VAR (aka

dynamical O.I.) assimilation for soil fields, dynamical O.I.) assimilation for soil fields, including an ensemble component => including an ensemble component => talk talk by J.-F. Mahfoufby J.-F. Mahfouf


Plans for 2007 and outlookPlans for 2007 and outlook► B matrix and background std dev.:B matrix and background std dev.:

Errors of the day for screeningErrors of the day for screening Gridpoint maps of Gridpoint maps of b’s for minimizationb’s for minimization Filtering of the ensemble bg errors (low-pass)Filtering of the ensemble bg errors (low-pass) Wavelets (A. Deckmyn; T. Landelius; L. Berre)Wavelets (A. Deckmyn; T. Landelius; L. Berre)

► Algorithms:Algorithms: 4D-VAR in a nutshell4D-VAR in a nutshell Some kick-off on a simplified microphysics scheme Some kick-off on a simplified microphysics scheme

for the mesoscale ?for the mesoscale ? Towards an integrated « Ensemble/Variational » data Towards an integrated « Ensemble/Variational » data

assimilation system; ETKF => Hirlamassimilation system; ETKF => Hirlam


Plans for 2007 and outlookPlans for 2007 and outlook► Observations:Observations:

SEVIRI radiances in Hungary and Morocco ?SEVIRI radiances in Hungary and Morocco ? SEVIRI/CSR monitoring and bias correction; surface SEVIRI/CSR monitoring and bias correction; surface

emissivity for IR SEVIRI channels (M. Stengel, Hirlam)emissivity for IR SEVIRI channels (M. Stengel, Hirlam) Radar radial winds in the Aladin-FR and Arome assim.Radar radial winds in the Aladin-FR and Arome assim. Continuation on radar reflectivity obs. op. & retrieval Continuation on radar reflectivity obs. op. & retrieval

methods (M. Juramethods (M. Juraššek, LACE; G. Haase, Hirlam)ek, LACE; G. Haase, Hirlam)► Operations:Operations:

Aladin-France: IDFI+radar winds+retuned Aladin-France: IDFI+radar winds+retuned b’s+60 levelsb’s+60 levels Arome RUC: 3h frequency, 4 times a day, 30h range, incl. Arome RUC: 3h frequency, 4 times a day, 30h range, incl.

Radar winds and reflectivities, 2.5 km over FranceRadar winds and reflectivities, 2.5 km over France Aladin-Hungary: T2m+RH2m+SEVIRIAladin-Hungary: T2m+RH2m+SEVIRI Aladin-Morocco: 3D-VAR plus NOAA and MSG radiancesAladin-Morocco: 3D-VAR plus NOAA and MSG radiances

Thank you for your Thank you for your attentionattention


Ludovic Auger (Aladin-FR 3D-VAR)Martin Bellus (DF Blending)Loïk Berre (new humidity control, ensemble B)Gergely Bölöni (ensemble B)Bernard Chapnik (FGAT, a posteriori diagnostics)Alex Deckmyn (wavelet Jb)Máriá Derková (DF blending)Rachida El Ouaraini (new humidity control)Claudia Faccani (radar radial winds)Claude Fischer (Aladin-FR 3D-VAR)Günther Haase (radar reflectivities)Marian Jurasek (radar reflectivities)Fatima Karbou (radiances over land)Sándor Kertész (FGAT, RUC, ECMWF LBC)Michal Májek (SEVIRI)Thibaut Montmerle (SEVIRI, radar radial winds)Roger Randriamampianina (SEVIRI)Benedikt Strajnar (RUC)Alena Trojaková (3DVAR in CZ, SEVIRI)Eric Wattrelot (radar reflectivities) And we’re sorry for those we forgot …

Contributors Contributors


ALADIN Rapid Update Cycle

Reference: RUC6

U

RHU

(…in RUC6 one analysis uses the same amount of data as in RUC3 which means that we exlude data around 03, 09, 15, 21 UTC…)

Red: improvement for all variables (U and RHU shown as examples)

…the meaning of this test is not so practical…


Complex wavelets for LAM (A. Complex wavelets for LAM (A. Deckmyn)Deckmyn)

► Wavelets are (partially) localised in both grid point space and Wavelets are (partially) localised in both grid point space and Fourier space.Fourier space.

► Diagonalization of B in wavelet space can reproduce local variations Diagonalization of B in wavelet space can reproduce local variations in the structure functions and standard deviations.in the structure functions and standard deviations.

► Current work is focussing on reproducing 3D structure functions for Current work is focussing on reproducing 3D structure functions for different variables (Alex Deckmyn, Tomas Landelius, Loik Berre)different variables (Alex Deckmyn, Tomas Landelius, Loik Berre)

Standard deviations of Temperature errorat the lowest vertical level: from data (left)and from wavelet-B (right).


Resolution volum, ray path : standard refraction (4/3 Earth’radius)

zh

rd

eZ

r

Nd

d

model level

Assimilation of Reflectivity : Observation operator implemented in the 3DVar ALADIN

• Bi-linear interpolation of the simulated hydrometeors (T,q, qr, qs, qg) • Compute « radar reflectivity » on each model level

Backscattering cross section: Rayleigh (attenuation neglected)

..., 0

),().,()(snowrainj

dDrDNjrDjr

Microphysic Scheme in AROME

Diameter of particules

• Simulated Reflectivity factor in « beam volum bv»

Antenna’s radiation pattern: gaussian function for main lobe

(side lobes neglected)

)..).,().(log10 4 dddrfr(Zbv

e

Resolution volum, ray path: standard refraction (4/3 Earth’s radius)

Aladin consortium activities in data assimilation

Documents

dvar cycle

h cycle cycle setup

dvar assimilation cycle

utc assimilation

utc run

utc ecmwf lbcs

shmu domain df blending

ruc lbc